Alexandros Koumparoulis

@adamlin120 thanks for reporting the issue. Can you please try https://github.com/NVIDIA/NeMo/compare/main...akoumparouli/update_megatron_gpt_cont_training and let me know if that fixes the issue for you? thanks.

Hi, which one do you want to use the mistral-7b or the mixtral-7b? if you want to use the mistral-7b please use convert_mistral_hf_to_nemo.py instead of convert_mixtral_hf_to_nemo.py. Thanks.

- EP=2 NGPU=8 - EP=1 TP=1 NGPU=8 These also work correctly

Hi, thanks for reporting this, Can you retry without EP and report back whether this improves the speed? In addition, I would encourage trying different TP/PP configurations to determine the...

jenkins

jenkins

jenkins

jenkins

jenkins

jenkins